Bayesian hierarchical modelling of the geospatial distribution of fasciolosis in dairy cattle and the impact on production: Application to the main milk-producing region (Galicia) in Spain

Abstract

Fasciolosis caused by Fasciola hepatica is a common parasitic infection among cattle in many countries. Although infected adult cows rarely show overt clinical signs, milk production may be impaired. Thus, significant production losses may occur in dairy herds with a high prevalence of fasciolosis. In this study, Bayesian hierarchical modelling was used to estimate the geospatial distribution of dairy cattle fasciolosis and its impact on milk production. The study was conducted in Galicia, the main milk producing region in Spain and a geographically heterogeneous area. The aims were: 1) to model the geospatial distribution of fasciolosis in dairy herds in the study area, 2) to identify clusters of herds with a high prevalence of fasciolosis, and 3) to assess the effect of fasciolosis on milk yield and quality. A large number of dairy cattle farms (n = 4907), of which 1660 provided production records, were surveyed. Fasciola infection status was determined by applying the MM3-SERO ELISA test to bulk tank milk samples. A high probability of infection was predicted in several zones, particularly in the centre, northeast and southeast of Galicia. Conversely, the predicted probability was very low in some parts of the northwest of the region. Infections with high within-herd prevalence (> 25% lactating cows infected) predominated. High within-herd prevalence was associated with loss of milk production (−1.387 kg/cow/ day, on average). No association between Fasciola infection and either milk fat or protein content was observed. This study has generated the first maps of the spatial distribution of the probability of Fasciola infection in dairy cattle herds in Galicia. The maps presented here can be used for reference purposes, enabling the design of better targeted fasciolosis control programmes in the region. Use of Bayesian hierarchical statistical analysis enabled us to ascertain the uncertainty of the predictions and to account for the spatial autocorrelation in the data. It also enabled us to generate maps showing the residual spatial variation in milk production, a topic that may deserve more detailed study.